One of the interesting things about the current debate over whether we’re about to go into a recession or not is the multitude of indicators that different people glom onto — without any expressed formal rationale for picking one over the other. See this list of people in the recession camp, here.
That’s why I found this (pre-pandemic) systematic comparison of the predictive content of indicators, by David Kelley of interest. The key figure is reproduced below.
The Conference Board’s Leading Economic Index is best at very short horizons. A literal reading of the August reading indicates we’ve been in a recession for a while. That being said, the 10yr-Fed funds spread is one of the components of the index, and one of the main drivers of the change over the last six months. If one is skeptical of this spread, one might be skeptical of this reading (the other main drivers have been consumer expectations of business conditions, and ISM new orders).
The 10yr-3mo spread (used in this post) is best at horizons of a year.
From the conclusion:
The results of this article show that at horizons roughly one year ahead and longer, the long-term Treasury yield spread has historically been the most accurate available “predictor” of recessions. That said, leading indexes have been better than individual leading indicators or financial data at signaling recessions in the near term. The ROC threshold indexes constructed here have also performed well as recession predictors in the near term because they are also effectively leading indexes that combine the information in the inputs to provide a more accurate measurement of coming economic activity.
Note these are predictors. They’re not indicators of whether we’re in a recession (e.g., Sahm rule).